Which brain-computer interface startups are furthest ahead?

Why has the brain-computer interface race suddenly become much more serious?

The BCI startup race has moved from isolated demonstrations into simultaneous clinical and commercial programs, so identifying who is actually ahead matters much more now.

Neuralink illustrates the change in scale. Its first human received an implant in early 2024. Two years later, the company said 21 people had joined its international clinical programs, up from three at the beginning of 2025. Neuralink has also moved beyond cursor demonstrations: participants have controlled computers for hours at home, used robotic arms, studied at university, created artwork and tested imagined-finger typing. Clinical programs multiplying since early 2025 is a big change, and we lay out how the race has changed over the past year in our neurotech report.

The competitive field has moved at the same time. Synchron accumulated a ten-person chronic implantation history across Australia and the United States before starting newer programs around device control. Precision Neuroscience says its Layer 7 cortical array has now been placed in more than 100 patients, mostly temporarily during surgery. Paradromics implanted the first participant in its long-term Connect-One study and subsequently demonstrated open-ended speech generation. NeuroXess has implanted fully wireless systems in people with paralysis in China.

Regulation is moving too. Precision received U.S. 510(k) clearance for its cortical electrode array for implantation of up to 30 days. Science Corporation received European approval for PRIMA, its retinal vision-restoration system. China's Neuracle Medical received approval to commercially use its implantable NEO system in people with spinal-cord injuries.

Several companies have therefore already proved that implanted interfaces can work in humans. The harder question now is which one can make that work reliably outside a research lab.

What does “furthest ahead” actually mean in brain-computer interfaces?

The BCI startup furthest ahead should be the company that has removed the most important barriers between an experimental neural interface and repeatable real-world use.

Those barriers include useful human performance, implant durability, surgical burden, regulatory progress and the ability to deploy the system repeatedly. Different companies lead on different parts of that journey.

Precision can say that more than 100 people have received its Layer 7 array, but most were temporary placements during neurosurgical procedures rather than long-term users living with a BCI. Synchron has implanted far fewer people, yet its system can be delivered through a blood vessel without opening the skull. Neuralink uses a more invasive penetrating interface but has produced richer computer control from people using the implant at home.

Neuracle creates another problem for any simple ranking. Its NEO system has already crossed from experimental device to commercially authorized medical product in China. That puts it ahead on commercialization. But its eight-electrode epidural interface focuses on restoring hand function through a robotic glove rather than providing the broad digital control Neuralink is pursuing.

Science Corporation presents a similar case. PRIMA has regulatory approval in Europe and restores limited central vision, but it is a retinal prosthesis rather than a general-purpose cortical interface.

We therefore have to distinguish high-bandwidth computer control, minimally invasive chronic implantation, commercial deployment and specialized restoration such as vision or hand movement.

Company	Strongest current evidence	Main limitation
Neuralink	High-performance chronic BCI used in everyday life by a growing participant cohort	Penetrating brain implant and specialized robotic surgery
Synchron	Chronic BCI implanted through the vasculature	Much lower neural-channel density
Neuracle Medical	First commercially approved invasive BCI and commercial implantation	Narrow hand-rehabilitation application
Precision Neuroscience	100+ human placements and FDA-cleared cortical array	Most placements temporary; full wireless BCI remains investigational
Paradromics	Fully implanted high-bandwidth BCI producing open-ended speech	Long-term clinical evidence currently starts with very few patients
Science Corporation	European approval and commercialization of PRIMA	Retinal prosthesis rather than general-purpose cortical BCI
NeuroXess	Fully implanted wireless cortical system performing real-world tasks	Very small chronic patient cohort
Blackrock Neurotech	Decades of human intracortical BCI evidence	Research platform remains ahead of its commercial implant program

Is Neuralink actually leading, or does it simply get the most attention?

Neuralink is currently the strongest overall general-purpose implanted BCI startup, and its lead now rests on chronic use and real-world performance rather than publicity.

According to Neuralink's two-year Telepathy review, its clinical cohort had grown to 21 participants by early 2026. Three participants alone had accumulated more than 4,900 hours of Telepathy use by early 2025, before the cohort expanded much further.

Performance has also moved beyond staged demos. Neuralink reports that several participants reached roughly the same 8–10 bits-per-second range that able-bodied people achieve with a physical mouse, with one participant exceeding 10 bits per second during his first week. The company later reported an 11.32-bit-per-second record using neural models pretrained on roughly 50,000 hours of unlabeled neural data.

More importantly, participants are using the system for ordinary activities. One medical student has reportedly operated it for as much as 17 hours in a day while studying. Another participant uses it for college coursework. Others have played games, created artwork and controlled robotic arms. Neuralink has also demonstrated imagined-finger typing at up to approximately 40 words per minute and started a separate speech program.

The company has had setbacks too. Its first participant experienced substantial retraction of implant threads, reducing the number of usable electrodes. Neuralink changed the procedure afterward and says 18 of the following 20 participants showed higher signal quality.

No direct competitor currently combines this level of chronic use, digital performance, breadth of applications and repeated implantation.

Has Neuralink solved the problem of keeping a high-bandwidth brain implant working for months?

Neuralink has shown that a high-bandwidth intracortical BCI can stay useful for months, but long-term durability is still one of the biggest questions around the platform.

The first Neuralink participant exposed the problem clearly. Some of the flexible threads inserted into the cortex retracted after surgery. The participant could continue using the system because Neuralink changed its software decoder to get more value from the remaining electrodes.

Later results are more encouraging. Neuralink says procedural changes improved signal quality across 18 of the next 20 participants. The company is also moving from approximately 1,000 electrodes toward a planned 3,000-electrode design while changing the mechanics of the implant to improve thread retention.

Its machine-learning work attacks another long-term problem: neural signals drift as tissue, electrodes and behavior change. Neuralink has reported that self-supervised models trained across tens of thousands of hours of neural recordings can help maintain decoder performance for weeks without constant recalibration.

That is strong evidence for months-long practical use. What we still do not have is the many-year reliability expected from mature implanted medical devices.

Is Synchron's easier surgery enough to make its BCI easier to scale?

Synchron currently has the easiest implantation strategy among leading chronic BCIs, although the simpler procedure comes with a clear bandwidth trade-off.

The Stentrode is delivered through the jugular vein and positioned inside a blood vessel next to the motor cortex. Synchron says the procedure takes roughly two hours, avoids cutting through the skull and normally allows participants to leave hospital the next day. A device delivered through a vein is just one answer to the surgery problem; we look at which technologies are gaining traction in our neurotech market report.

That is a very different proposition from Neuralink and Paradromics, which place electrodes into cortical tissue, or surface-array systems that still require direct access to the brain. It also means Synchron can potentially rely on specialists already trained in neurointerventional catheter procedures rather than building an entirely new surgical workflow.

Synchron has accumulated relatively long human experience too. Its early Australian SWITCH study preceded the six-person U.S. COMMAND study, giving the platform around ten historically reported chronic recipients before newer programs began. Users have demonstrated texting, web browsing and digital-device control from home.

The cost of that simplicity is neural resolution. A vascular electrode sits farther from individual neurons than an intracortical array, so Synchron captures much less detailed neural information. That could limit continuous cursor control, complex robotic movement and direct speech decoding.

For many patients, maximum bandwidth may be unnecessary. Choosing icons, navigating menus, writing messages and operating a tablet can already restore meaningful independence. Synchron's work with Apple's BCI Human Interface Device protocol reinforces that approach because consumer software can be adapted around a smaller set of reliable neural commands.

Synchron is betting that useful control with a much easier procedure will beat maximum neural bandwidth in a large part of the market. That remains unproven, but today it gives the company the clearest path to scaling chronic implantation through existing hospital infrastructure.

Does Precision Neuroscience have the best compromise between bandwidth and brain surgery?

Precision Neuroscience may have the most attractive middle-ground architecture in the BCI race, but it still needs to prove that advantage in people living chronically with the complete system.

Its Layer 7 Cortical Interface received FDA 510(k) clearance for recording, monitoring and stimulating electrical activity on the brain surface for up to 30 days. Each ultra-thin array can contain 1,024 electrodes positioned directly on the cortex without inserting them into brain tissue.

Precision also reports more than 100 patient placements across more than 18 hospital partners and has demonstrated procedures using multiple arrays totaling 4,096 electrodes.

Those numbers sound stronger than they really are if we compare them directly with Neuralink. Most Precision placements have been temporary during neurosurgical procedures and the arrays were removed afterward. They provide strong evidence about surgical handling, recording quality and hospital adoption, but much less evidence about months of home use, wireless reliability or long-term signal drift.

The architecture itself remains compelling. Synchron avoids brain surgery but records from inside a blood vessel. Neuralink and Paradromics get much closer to individual neurons but penetrate cortical tissue. Precision sits directly on the cortical surface and may capture considerably richer information than a vascular BCI without using penetrating electrodes.

Surface recordings still sacrifice some spatial detail compared with intracortical signals, especially for very fine motor control or high-speed speech decoding.

For now, Precision has one of the strongest technical compromises in the industry. Its missing proof is chronic use of the finished wireless system.

Approach	Representative startup	Main advantage	Main unresolved trade-off
Intracortical flexible threads	Neuralink	High-bandwidth neural signals	Tissue penetration and complex surgery
Intracortical microelectrode array	Paradromics	High neural data rate, particularly for speech	Invasive implantation and limited chronic human data
Cortical surface array	Precision Neuroscience	High electrode density without penetrating brain tissue	Lower spatial resolution than intracortical recording
Endovascular electrodes	Synchron	No open brain surgery	Lower signal resolution and channel count
Epidural surface interface	Neuracle	Simpler implantation and commercial rehabilitation use	Narrower control bandwidth
Subdural flexible interface	NeuroXess	Richer cortical access without penetrating neurons	Very early clinical scale
Retinal photovoltaic implant	Science Corporation	Already approved for a specific functional outcome	Specialized vision prosthesis, not general-purpose BCI

Has Paradromics caught Neuralink in the race to restore speech?

Paradromics has become one of the most credible speech-BCI challengers to Neuralink, although its clinical program is still much smaller.

The company received FDA authorization for its Connect-One early feasibility study in late 2025 and implanted its first long-term participant at the University of Michigan the following year. That followed a temporary first-in-human recording during epilepsy surgery in 2025.

Paradromics then reported that its first chronic participant, a woman with motor-neuron disease, could use Connexus to produce her own words in real time, answer open-ended questions and communicate with family during a phone call. The system converted neural activity into text and synthesized speech rather than simply selecting from predefined phrases.

Speech may be where high-bandwidth intracortical recording has the clearest medical advantage. A person with ALS can retain normal language ability while losing the muscles needed to speak, making attempted-speech signals especially valuable.

Paradromics designed Connexus around that problem. Its intracortical arrays record individual-neuron activity and send the information through an implanted transceiver to external hardware.

The biggest weakness is sample size. Neuralink already has a much larger implanted cohort and its own dedicated speech program. Paradromics now has a compelling result; it still has to show that the result can be repeated across more patients.

Is Neuracle Medical actually ahead of every U.S. BCI startup?

Neuracle Medical is ahead of every U.S. BCI startup on one concrete milestone: it has obtained commercial regulatory approval for an invasive brain-computer interface.

Chinese regulators approved the company's NEO system for people with quadriplegia caused by cervical spinal-cord injury. Nature Biotechnology described NEO as the world's first invasive BCI approved for commercial use.

NEO is a coin-sized wireless implant with eight electrodes positioned above the brain rather than penetrating neural tissue. The system decodes attempted hand movement and uses those signals to operate a pneumatic robotic glove.

Its supporting clinical program involved 36 participants, according to reporting around the approval, with improvements in hand-grasp function reported across the study population. The first commercial implantation subsequently took place at Huashan Hospital.

NEO is far less ambitious as a general neural interface than Neuralink or Paradromics. Eight epidural electrodes cannot provide the same amount of neural information as systems with hundreds or thousands of cortical contacts.

But Neuracle has already turned its narrower technology into an approved treatment. On commercial readiness, it is ahead.

Is China catching the U.S. faster than most BCI rankings suggest?

China is already a first-tier BCI competitor, especially when we look at clinical deployment rather than media attention.

Neuracle's approval is the clearest example, but NeuroXess is pursuing a technically more ambitious route. The Shanghai company has implanted a fully implanted, fully wireless flexible cortical BCI in people with severe paralysis. Neuracle's approval and NeuroXess's wireless implant show challengers closing in, and we look at how far the challengers are behind in our neurotech report.

Its first long-term participant has used the system for practical digital activities and writing. A second participant reportedly combined BCI control with functional electrical stimulation to perform tasks including eating and drawing. NeuroXess says its end-to-end latency has fallen to around 50 milliseconds.

The company has also moved into a formal registration clinical trial and entered China's special-review pathway for innovative medical devices.

Some Chinese provinces have started creating medical-service pricing mechanisms for BCI procedures, which could help solve the reimbursement problem before patient volumes become large.

The U.S. still has several of the strongest high-bandwidth systems, particularly Neuralink, Paradromics and Precision. China, however, is already ahead on some parts of clinical commercialization.

Where does Blackrock Neurotech fit after more than two decades of human BCI research?

Blackrock Neurotech remains one of the most scientifically proven intracortical BCI companies, but it has fallen behind newer startups in turning that research history into a modern commercial implant.

The Utah Array behind much of Blackrock's technology has been implanted in people since 2004. Systems built around it have enabled cursor control, robotic-arm manipulation, handwriting decoding, speech research and artificial sensory feedback through programs such as BrainGate.

Industry datasets count roughly 40 people implanted with Blackrock hardware across research programs, giving the company more historical human experience than most commercial BCI startups.

The traditional system architecture is the problem. Many Utah Array research systems rely on a connector passing through the skull, which creates infection, maintenance and everyday-use problems.

Blackrock's MoveAgain program and newer fully implantable designs are supposed to remove that limitation. MoveAgain has FDA Breakthrough Device designation, but commercialization has progressed more slowly than the company's scientific lead would suggest.

Today, Blackrock remains a research heavyweight while Neuralink, Synchron and Precision look further ahead in productizing the technology.

Does Science Corporation count as a BCI leader now that PRIMA can be sold in Europe?

Science Corporation is one of the most commercially advanced neurotechnology startups today, although PRIMA belongs to a different category from Neuralink-style general-purpose BCIs.

PRIMA replaces part of the eye's lost photoreceptor function. A tiny photovoltaic implant sits beneath the retina while camera-equipped glasses capture visual information and project it onto the implant with near-infrared light. Remaining retinal neurons carry that information toward the brain.

In the European study supporting PRIMA, 84% of 38 patients reportedly regained the ability to read letters and numbers. Individual patients have used the system for activities including reading text and completing visual puzzles.

PRIMA has now obtained European authorization for geographic atrophy caused by dry age-related macular degeneration, and Science is activating commercial sites and reimbursement pathways.

Science is also developing broader biohybrid neural interfaces, but those programs remain much earlier. PRIMA makes Science a leader in commercial neural prosthetics, not in general-purpose cortical communication.

Who is winning on raw neural bandwidth?

Neuralink and Paradromics currently have the strongest position when we focus on extracting large amounts of information directly from cortical activity.

Neuralink's current generation uses roughly 1,000 electrodes distributed across flexible threads inserted into the cortex, and its next iteration is intended to move toward 3,000. More importantly, participants have translated those signals into information-transfer rates around or above conventional mouse performance.

Paradromics uses hundreds of intracortical electrodes in Connexus and has focused heavily on the amount of useful neural information the implant can transmit. Its speech results suggest that the system captures enough detail to distinguish patterns associated with listening, preparing speech, attempted speech and imagined speech.

Precision can place 1,024 electrodes on each cortical surface array and has demonstrated procedures using four arrays, reaching 4,096 contacts. Those electrodes, however, record broader cortical field activity rather than individual-neuron spikes.

Synchron operates at the other end of the spectrum with a much smaller number of electrodes in a blood vessel.

Raw electrode counts alone cannot tell us who has the best BCI because signal quality, electrode placement and decoder stability matter just as much. For demanding applications such as fast speech or detailed motor control, though, intracortical systems currently have the clearest advantage.

Does having more BCI patients actually tell us who is ahead?

Patient count is useful for filtering out BCI hype, but the number becomes misleading if we mix temporary surgical placements, chronic implants and commercially treated patients.

Precision provides the clearest example. Its more than 100 human placements give it the largest recent exposure among high-density cortical-interface startups, but many arrays were placed during another neurosurgical procedure and removed afterward.

Neuralink's cohort is smaller but consists of long-term implanted participants who repeatedly use the system. That produces evidence about battery life, wireless operation, thread stability, decoder drift, home support and daily behavior that an operating-room placement cannot provide. Months with an implant at home expose real hardware problems, and we look into the biggest unsolved ones in our report on who's ahead in neurotech.

Synchron has fewer patients again, but its chronic implants have produced multi-year experience.

Paradromics is much earlier. Its first permanent Connect-One participant has generated impressive speech results, but one chronic user tells us little about population-wide reliability.

Neuracle sits in a different position: its 36-person supporting clinical program was strong enough to lead to regulatory approval and commercial treatment.

The useful comparison is how many people have used something close to the intended final product, and for how long.

Which BCI startups are closest to treating large numbers of people?

Neuracle Medical is closest to ordinary commercial treatment today, Synchron has the simplest path toward larger chronic implantation volumes in the West, and Neuralink has the strongest route toward scaling a much higher-bandwidth system.

Neuracle has already crossed into commercial medicine in China.

Science Corporation has also crossed the approval line in Europe with PRIMA, although its product addresses retinal vision loss rather than general-purpose computer control.

Among U.S. cortical communication systems, Synchron's procedure is easiest to imagine spreading through existing hospitals. If larger trials confirm safety and benefit, neurointerventional specialists could perform the implantation without hospitals buying an entirely new surgical platform.

Neuralink's route is harder but potentially more powerful. Its robot, implant and hospital workflow all need to become highly standardized. If that happens, automation could eventually make delicate cortical implantation reproducible across many centers.

Precision could still emerge as the strongest compromise because it already works with a broad hospital network and uses a less aggressive interface than penetrating systems while retaining much higher electrode density than vascular BCIs.

Paradromics remains earlier in that process. Its strongest chance is to show that outcomes such as natural speech are valuable enough to justify the extra surgical burden.

Nobody has demonstrated anything close to mass BCI deployment yet. Even the leaders still operate at patient counts measured in tens or, under broader definitions, low hundreds.

So which brain-computer interface startups are furthest ahead?

Neuralink is currently the furthest-ahead startup if we mean the company closest to a versatile, high-performance brain-computer interface that people can repeatedly use in everyday life.

We would put Neuralink first overall because it has combined four things competitors usually possess separately: a growing chronically implanted cohort, high-bandwidth digital control, substantial real-world use and an increasingly repeatable surgical process. Its participants have moved beyond laboratory cursor tests into education, creative work, robotic control and long daily sessions.

Synchron has the strongest alternative strategy. Its lower-bandwidth interface has a much easier surgical path and years of chronic human evidence. If reliable communication turns out to matter more commercially than extracting as much neural information as possible, Synchron could scale faster.

Neuracle Medical belongs near the top because it has already achieved something the U.S. leaders have not: commercial authorization of an invasive BCI. NEO is narrower and lower-bandwidth, but it is already a medical product rather than only an experimental platform.

Precision Neuroscience is probably the most dangerous medium-term challenger. More than 100 human placements, a cleared cortical array, a large hospital network and a surface-electrode architecture give it a strong base. Its biggest missing piece is extensive chronic use of the finished wireless system.

Paradromics has made the sharpest recent move upward. Its first long-term participant generated open-ended real-time speech, giving the company meaningful evidence in one of the most valuable BCI applications. We now need to see that performance repeated across more people.

Science Corporation is already more commercial than most of the field through PRIMA, but its core approved product belongs to a different category. NeuroXess is becoming an important Chinese challenger. Blackrock Neurotech still has exceptional scientific credibility but has moved more slowly toward a modern fully implanted commercial system.

Position	Startup	Where it is furthest ahead	Our current judgment
1	Neuralink	General-purpose high-bandwidth chronic BCI	Strongest overall lead
2	Synchron	Minimally invasive chronic computer control	Best scalability architecture today
3	Neuracle Medical	Regulatory approval and commercial invasive BCI use	Commercial leader
4	Precision Neuroscience	Surface-interface clinical scale and U.S. regulatory progress	Strongest potential compromise
5	Paradromics	High-bandwidth speech restoration	Fastest-rising direct challenger
6	Science Corporation	Approved neural prosthesis and commercial execution	Leader in vision, not general-purpose BCI
7	NeuroXess	Fully implanted wireless cortical BCI in China	Important emerging challenger
8	Blackrock Neurotech	Long-term intracortical scientific evidence	Research leader, commercial laggard